Applying a marketing attribution model
Marketing attribution, fractional attribution, or multi-touch attribution (MTA) are terms defining the same concept:
"The process of identifying a set of user touch points that contribute in some manner to a desired outcome, and then assigning a value to each of these events."
The underlying methodology and algorithm, however, are not universally agreed upon. Let's go through the various options, their benefits, and short-comings.
There are multiple types of attribution models. An organization needs to decide which model to utilize and even within a model, there are more options which reduce the consistency.
There are many models. Here are a few to consider:
First-touch: Some organizations use first touch, wherein the first marketing contact point would receive 100% of the credit. The rationale is that that first contact point is what lured the customer into the funnel, and the rest was just an inevitable journey toward conversion. This simple model is problematic because they ignore most of your marketing efforts.
Last-touch: The simplest model. All of the credit goes to the touch that happened just before the conversion event. The thinking here is that this is what convinced the person to actually pull the trigger and make the purchase, so that’s what we want to credit.
Equal distribution: Also called linear distribution is very egalitarian; it just splits credit evenly across all of the channels. This model is easy to understand and an improvement on last and first touch, because it doesn’t ignore any of your channels.
Position-weighted: Sometimes called a U-shaped model, last and first touch are deemed the most important, but you don’t give them all the credit. You divide some of it between your middle touches. Here’s where we start to get subjective, because unlike the first three models, we have to make decisions about how to apportion this credit. How do we know that 40% of the credit is appropriate for the first and last touch? We don’t.
Decay model: More recent touch-points get the most credit, and as we move backward in time, our channels get less credit. This has some intuitive appeal, but again, it’s subjective: how do you determine which decay function to use? Should it be linear, exponential, should it be based upon the actual date of the touch-point?
Algorithmic: This is the most accurate and most difficult. Using machine learning, all touch-points and the attributes and metadata associated with them (frequency, time of day, creative, publisher, etc...) that lead to a conversion and those which do not lead to a conversion (negative outcomes) are weighted. Since it uses your organization's data, it's the most accurate.
Since there are multiple ways to apply attribution, the concept is not universally accepted. Some organizations are skeptical in how marketing is measured and the credit attributed to each touch point marketing claims.
Aligning with Finance can help if the team is willing to dig deep into how marketing efforts are being tagged and tracked, and understanding the methodology for assigning credit.
Implementing a solution
Once you've gotten buy-in, adoption is another hurdle. It requires metrics collection discipline and governance. If each marketing tactic is tracked and collected in separate systems, assigning credit will be impossible.
You can leverage an ad server, utilize an attribution modeling company's third-party pixel to piggy back on all your tactics, or a newer breed of data experience platform provides this unified tracking platform with the governance built-in.
To outsource or not
There are third parties that provide an out-of-the-box solution. Gartner has provided their assessment of the landscape in their Market Guide for Attribution and Marketing Mix Modeling.
While a huge undertaking, if you have a robust data science discipline within your organization, it might be worth evaluating whether your internal resources have the skills and cycles to undertake the effort themselves.
Whatever route you take, you'll find that implementing a marketing attribution solution will aid in better understanding how your marketing affects the bottom line and can aid in budget allocations and decisions.